Stored platelet hemostatic phenotype and function is not altered when donors are on testosterone replacement therapy

Abstract

Background:

Critical shortages in the national blood supply have led to a re-evaluation of previously overlooked donor sources for blood products. As a part of that effort, red blood cells collected from therapeutic phlebotomy of donors on testosterone replacement therapy (TRT) have been conditionally approved for transfusion. However, platelets from TRT donors are not currently approved for use due to limited data on effects of supraphysiologic testosterone on recipient safety and platelet function. The objective of this study was to provide a comprehensive profile of phenotype and function in platelets from TRT and control donors.

Study Design and Methods:

Platelets in plasma were collected from TRT and control donors (N = 10 per group; age- and sex-matched) and stored at room temperature for 7 days. On storage Day 1 (D1) and Day 7 (D7), platelet products were analyzed for platelet count, metabolic parameters (i.e., glucose, lactate, mitochondrial function), surface receptor expression, aggregation, thrombin generation, and thrombus formation under physiological flow conditions.

Results:

TRT donor platelets were not significantly different than control donor platelets in terms of count, surface phenotype, metabolic function, ability to aggregate, thrombin generation, or ability to form occlusive thrombus under arterial flow regimes. Both groups were similar to each other by D7, but had significantly lost hemostatic function compared to D1.

Discussion:

Platelets derived from donors undergoing TRT have similar phenotypic and functional profiles compared to those derived from control donors. This suggests that therapeutic phlebotomy of TRT donors may provide a useful source for platelet products.

1 |. INTRODUCTION

Forecasts for a robust platelet product supply in the United States (US) in the coming years are bleak due to combinations of an aging donor population,¹ increased distribution and outdating,² and the looming fears of another pandemic.^3,4 As such, alternative strategies to increase the available platelet supply are an important area of study. These strategies can be seen in changes to current transfusion practices, and blood product inventory management and distribution logistics.⁵ They can also be donor centric, through marketing to younger demographics and the controversial concept of donor remuneration. Lastly, other mechanisms are product centric, such as (i) lowering platelet counts per unit from 3 to 2.5 × 10¹¹, (ii) producing whole blood-derived platelets (common outside of the US), and (iii) extending apheresis shelf life through altered storage temperatures (e.g., cold and/or frozen storage) and storage solutions (e.g., platelet additive solutions to minimize metabolic activity).

Interestingly, the average demographic for apheresis platelet donors in the US are males greater than 50 years of age^6,7—the same demographic that receives a majority of prescriptions for testosterone replacement therapy (TRT).^8–10 Known side effects of TRT are polycythemia and erythrocytosis,^11–13 which lead to the need for therapeutic phlebotomies to reduce a patient’s hematocrit (HCT), thereby minimizing blood viscosity and the associated increased risk of thromboembolism,¹⁴ and major adverse cardiac events.¹⁵ HCT thresholds indicative of therapeutic phlebotomy are not standardized, with reports of HCT values anywhere between 50% and 55% triggering a phlebotomy prescription.^11,16 Importantly, the US Food and Drug Administration (FDA) allows red blood cell (RBC) collections from therapeutic phlebotomy of TRT donors to be used for transfusion, but not plasma and platelet components.^17,18 This guideline stems from the limited information we have regarding the effects of testosterone in patients receiving transfusions (to include androgen-sensitive patients such as neonates) containing large plasma volumes (i.e., plasma and platelet components), as well as the direct effects of testosterone on platelet storage and function.¹⁹

In a recent literature review of the effects of exogenous androgens on platelet function, Rosca et al. report that data on the direct effects of testosterone on platelet function are still scarce and conflicting,²⁰ despite 50 years of study. Exogenous testosterone has been found to be both activating²¹ and inhibitory²² regarding healthy donor platelet aggregation. In older healthy donors, high levels of endogenous testosterone are associated with decreased aggregation responses,²³ whereas low levels of plasma testosterone (incurred via castration) were also found to be associated with reduced platelet aggregation.²⁴ Moreover, there have been few studies on platelet function in response to TRT. A randomized placebo-controlled study found that testosterone cypionate administration resulted in increased TXA₂ receptor expression on the platelet surface, which was associated with a ~10% increase in TXA₂-induced aggregation, but no differences in thrombin induced aggregation.²⁵ Thus, there remains limited and unclear evidence on the effects of TRT on platelet function.

The aim of our study was to determine if platelet products (PLT) derived from donors on TRT were significantly different with respect to hemostatic function compared to PLT derived from control donors (CD). In contrast to most studies using platelet-rich plasma (PRP) generated from donors and tested immediately using aggregation (as outlined above) as an indicator of platelet function, we took a comprehensive approach to measure phenotype and functional metrics associated with primary and secondary hemostatic function, in both static and dynamic settings. Furthermore, we assessed the intrinsic hemostatic capacity of PLT over time—by assaying PLT on both Day 1 (D1) and Day 7 (D7) of room temperature (RT) storage.

2 |. METHODS

2.1 |. Donors and blood collection

Donors currently on TRT and with prescriptions for therapeutic phlebotomy (N = 10, “TRT”) and 10 additional healthy volunteers (CD, age- and sex-matched to TRT donors) were recruited for this study. All donations were obtained in accordance with our approved Institute Review Board protocol (#20200109). Inclusion and exclusion criteria can be found in Table S1. Donor demographics, (ethnicity [self-assigned], blood type, age), vitals (blood pressure, pulse, temperature, weight, height), and hemoglobin levels were recorded before donation. For TRT donors, the type of therapy and dose were recorded when available (Table S2), and for all donors, additional medications were recorded when available (Table S3). For each donor, a small volume of whole blood was collected into a BD Vacutainer Serum Tube and a BD Vacutainer EDTA Tube (BD, San Diego, CA, USA), followed by collection of a unit of whole blood using the IMUFLEX^™ WB-SP Blood Bag System (TerumoBCT, Lakewood, CO, USA).

2.2 |. Serum isolation from whole blood & testosterone measurement

Whole blood in serum tubes was mixed evenly by inverting at the time of collection and placed upright at 22°C for 30–60 min to allow for clotting. After incubation, tubes were centrifuged (1500 g, 10 min, 22°C), then supernatant (serum) gently removed and frozen in 500 μL aliquots at −80°C. Frozen serum aliquots were sent to University of Colorado Health Laboratories for testosterone assessment (Test # LAB173) via ARUP (Test code #2004246) for both total testosterone by mass spectrometry (ng/dL; bound to SHBG or albumin) and free testosterone by dialysis (pg/mL; bioavailable) measurements.

2.3 |. Platelet isolation and storage

PLTs were prepared according to Method 6–12 from the AABB Technical manual 20th Edition.²⁶ Full details can be found in the Supplementary Materials. In brief, whole blood was collected and platelet concentrates derived using the PRP method. After overnight incubation, platelet concentrates were evenly split into two mini bags (~30 mL each, proprietary small volume platelet storage bags, validated in-house). One bag was designated Day 1, or “D1,” and assayed immediately, whereas the other bag was stored for an additional 6 days (22°C with agitation) and used on Day 7 (“D7”) for assays.

2.4 |. Complete blood count and arterial blood gases

Complete blood count on whole blood collections (EDTA tubes at time of donation) and on PLT (250 μL) were performed using both Sysmex XN1000BB and XE-2100D platforms (Siemens, Munich, Germany). Data are reported as platelet count (PLT, ×10³/μL; Table S4), mean platelet volume (MPV, fL), RBC count (10⁶/μL), hematocrit (HCT, %), hemoglobin (HGB, g/dL), and mean corpuscular volume (MCV, fL). Arterial blood gas measurements on PLT (85 μL) were performed using an ABL90 FLEX PLUS blood gas analyzer (Radiometer America, Brea, CA, USA), and data reported as pH, bicarbonate (HCO₃₋, mmol/L), partial pressure of oxygen (pO₂, mmHg), glucose (Glu, mmol/L), and lactate (Lac, mmol/L).

2.5 |. Flow cytometry

Surface receptor phenotyping was performed by staining 10 μL PLT with an antibody cocktail containing, CD41, CD42b, CD49b, CD41/CD61, GPVI, CD62P, Lactadherin-FITC, and TruStain FcX. Antibody fluorophore and clone information can be found in Table S5. Samples were incubated with antibodies for 15 min 20–22°C, followed by fixation with 1 mL phosphate-buffered saline +1% paraformaldehyde at 2–6°C for ≥30 min prior to acquisition on a BD FACSLyric^™ (BD). Flow cytometry data were analyzed using FlowJo v10.8.1 software (BD).

2.6 |. Aggregometry

Aggregometry was performed using a light transmission aggregometry device (CHRONO-LOG^® Model 700 Whole Blood/Optical Lumi-Aggregometer, CHRONO-LOG, Havertown, PA, USA) per manufacturer’s protocol. In brief, 500 μL of PLT was mixed with either adenosine diphosphate (ADP, 10 μM) or collagen (COL, 5 μg/mL) to induce aggregation, and data collected for 10 min. Data recorded are aggregation over time (%/min), maximum amplitude (maximum % aggregation, “% Max Amp”), and area under the curve (AUC).

2.7 |. Thrombin generation

Thrombin generation was measured in PLT (80 μL) via calibrated automated thrombinoscope (CAT; Diagnostica Stago Inc., Maastricht, The Netherlands) per manufacturer’s protocol. Stago reagents used were: PRP Reagent, Thrombin Calibrator, and the FluCa Kit. Endogenous thrombin potential (ETP, nM × min), time to thrombin generation (Lagtime, min), maximal thrombin generation (Peak, nM), and time to peak (Time to Peak, min) were recorded.

2.8 |. Hemostatic function under flow

The intrinsic hemostatic capacity of PLT was measured under flow conditions using the Total Thrombus formation Analysis System (T-TAS, DiaPharma, West Chester Township, OH, USA) using the HD chip per manufacturer’s protocol. In brief, 20 μL calcium corn trypsin inhibitor reagent was gently mixed with 480 μL PLT, then 450 μL of this mixture transferred to the sample reservoir on the T-TAS. The filled reservoir was inverted and secured on the HD chip, then the assay started. The HD chip is coated with both collagen and thromboplastin (tissue factor) to activate both primary and secondary hemostasis, and flow is maintained at sheer rate of 1200/s, which mimics in vivo arterial blood flow. Data are reported as area under the curve (AUC), occlusion start time (OST, sec), occlusion time (OT, sec), and we generated a secondary variable to capture the growth time (GT, sec) by subtracting the OST from the OT.

2.9 |. Mitochondrial function assessment

Platelet mitochondrial function was assayed using a Sea-horse XFe24 device (Agilent, San Diego, CA, US). Detailed methods can be found in the Supplementary Materials. In brief, washed platelets were generated from PLT, and the oxygen consumption rate (OCR) was measured in response to subsequent stimulation with Oligomycin, carbonyl cyanide-4 phenylhydrazone (FCCP), and rotenone/antimycin A. OCR data are reported in pmol/min, with additional metrics generated by Agilent built macros: basal respiration, maximal respiration, spare respiratory capacity, ATP production, proton leak, and non-mitochondrial respiration.

2.10 |. Statistical analyses

Data were analyzed and visualized using GraphPad Prism version 10.0.1 (218) for Windows (GraphPad Software, Boston, MA, USA). For donor-specific continuous variables, data are reported as median (interquartile range [IQR]) and compared via unpaired t-test. For donor-specific categorical variables, data are reported as number (frequency) and were compared via Fisher’s Exact test. Data in figures are represented as individual points, with error bars reported as median (IQR). Kinetic data (i.e., aggregation curves, thrombin generation curves, T-TAS curves, and mitochondrial OCR curves) were represented as mean (standard error of the mean [SEM]). Comparison between groups (CD vs. TRT) of a variable at a specific time point (D1 or D7) was performed via nonparametric unpaired t-test (Mann–Whitney). Statistical significance is denoted by asterisks: *p < .05; **p < .01; ***p < .001; ****p < .0001. Comparison of a given variable over time (D1 vs. D7) between both groups was performed by two-way ANOVA (Fisher’s LSD test). p Values for D1 versus D7 in CD are denoted by pound symbols: ^#p < .05; ##p < .01; ^###p < .001; ^####p < .0001. p Values for D1 versus D7 in TRT are denoted by dollar symbols: ^$p < .05; $ $p < .01; ^{$ $ $}p < .001; ^{$ $ $ $}p < .0001.

3 |. RESULTS

Our two cohorts of control donors (“CD”) and donors on testosterone replacement therapy (“TRT”) were age- and sex-matched, however there were more donors with type O blood in the TRT group (CD, 1 [10%]; TRT, 6 [75%]; p = .013; Table 1). Vitals at donation were similar between groups, but TRT donors had increased hemoglobin (median 18.0 g/dL) compared to controls (median 16.2 g/dL; p = .022), despite similar MCV values (Figure S1). TRT therapy types are detailed in Table S1. As expected, both total and free testosterone in TRT sera were significantly elevated compared to CD sera (Figure S2).

TABLE 1.

Demographic and donation visit data.

Variable	CD N = 10	TRT N = 10	p Value
Age, y	56.5 (44.3–69)	62.5 (48–65.8)	.832
Ethnicity
American Indian or Alaska Native	0 (0)	0 (0)	>.999
Asian	0 (0)	0 (0)	>.999
Black	0 (0)	1 (10)	>.999
Native Hawaiian or Other Pacific Islander	0 (0)	0 (0)	>.999
White	9 (90)	9 (90)	>.999
Other	1 (10)	0 (0)	>.999
Hispanic	2 (20)	1 (10)	>.999
Blood type*
A	6 (60)	2 (25)	.188
B	2 (20)	0 (0)	.477
AB	1 (10)	0 (0)	>.999
O	1 (10)	6 (75)	.013
Hemoglobin (Hgb), g/dL	16.2 (15.3–16.6)	18 (16.8–18.8)	.022
Blood pressure, mmHg
Systolic	132.5 (126.5–139.5)	129 (126.3–135.8)	.835
Diastolic	83.5 (75–85)	81.5 (76–86.8)	.963
Pulse, beats per minute (bpm)	73 (72–81.8)	77 (70.8–85.5)	.770
Temperature, °F	97.7 (97.3–98.5)	98.1 (97.5–98.3)	.743
Weight, lbs	187.5 (180–203)	195 (182.8–222.5)	.231
Height, in	69.5 (69–71.5)	69.5 (68–71)	.889

Note: For continuous variables, data are reported as median (interquartile range) and compared using an unpaired t-test. For categorical variables, data are reported as number (frequency) and were compared using a Fisher’s Exact test. Statistically significant p values are bolded and italicized.

^*For TRT group, only 8 donors had blood type information available.

Abbreviations: CD, control donors; TRT, donors on testosterone replacement therapy.

Whole blood CBC at time of donation showed no significant difference in platelet counts (Figure 1A) or MPV (Figure 1B) between CD and TRT. Although erythrocytosis is a common adverse effect of testosterone therapy,²⁷ there was no increase in RBC count (Figure 1C). In contrast, the hematocrit (HCT) of TRT donors was elevated compared to controls, however levels remained below reported therapeutic phlebotomy cut-off values of ~54%²⁸ (Figure 1D). Despite elevated HCT in TRT donors, there was no associated increase in hemoglobin (HGB) concentration (Figure 1E). In PLT generated from TRT and CD donors, platelet counts were reduced by 40% in TRT compared to CD on D1, and this difference persisted through D7 (Figure 1F). MPV values increased in both CD and TRT PLT over storage (Figure 1G). There were no differences in PLT pH between CD and TRT on D1; however, pH decreased in both groups by D7, while still staying above the disqualification level for transfusion (pH < 6.2²⁹; Figure 1F).

FIGURE 1.

Complete blood counts in study groups. (A) Platelet count (PLT, 10³/μL), (B) mean platelet volume (MPV, fL; shaded region is normal range of 7.5–11.5 fL), (C) red blood cell count (RBC, 10⁶/μL), (D) hematocrit (HCT, %; shaded area indicates 50%–55%, the upper cut off indicating need for therapeutic phlebotomy^11,16), and (E) hemoglobin (HGB, g/dL) in whole blood of CD and TRT donors at time of donation. (F) Platelet count, (G) mean platelet volume (MPV, fL), and (H) pH (dashed line indicates lowest pH allowed for platelet products) in CD and TRT platelets on D1 and D7 of storage. Circles denote CD study group and triangles denote TRT study group. Data are visualized as median ± IQR, N = 10 per group. Data in A–E were analyzed using a Mann–Whitney test. Data in F–H were analyzed using a two-way ANOVA (Fisher’s LSD test). Statistical significance between CD and TRT on given day denoted as: *p < .05; **p < .01. Statistical significance between CD on D1 and D7 denoted as: ^#p < .05; ^###p < .001. Statistical significance between TRT on D1 and D7 denoted as: ^{$ $}p < .01; ^{$ $ $ $}p < .0001.

We next characterized stored platelet phenotype via flow cytometry. There was a statistically significant increase in CD41 GMFI between D1 CD (median [IQR], 6688 [5037–7977]) and TRT (8033 [6224–11,323]) (p = .027; Figure 2A). While there was no difference in CD41 GMFI between TRT and CD at D7, CD41 GMFI was reduced in TRT D7 compared to TRT D1. CD42b expression (GP1bα, part of the Von Willebrand Factor [VWF] receptor) was similar between groups at D1, but TRT PLT had slightly decreased CD42b expression by D7 compared to D1 (p = .047; Figure 2B). There were no differences across groups or time in CD49b expression (Integrin α2, binds to collagen; Figure 2C). GPVI expression was similar between CD and TRT at D1, but slightly reduced by D7 in both groups compared to D1 (Figure 2D).

FIGURE 2.

Platelet surface receptor phenotype by flow cytometry. Geometric mean fluorescent intensity (GMFI) of (A) CD41, (B) CD42B, (C) CD49B, (D) GPVI, (E) Lactadherin, (F) CD62P, (G) CD41/61, and (H) CD63. Circles denote CD study group and triangles denote TRT study group. Data are visualized as median ± IQR, N = 10 per group, and analyzed using two-way ANOVA (Fisher’s LSD test). Statistical significance between CD and TRT on given day denoted as: *p < .05. Statistical significance between CD on D1 and D7 denoted as: ^#p < .05; ^##p < .01; ^####p < .0001. Statistical significance between TRT on D1 and D7 denoted as: ^$p < .05; ^{$ $}p < .01; ^{$ $ $}p < .001; ^{$ $ $ $}p < .0001.

We measured markers associated with platelet membrane remodeling, activation, and granule mobilization. Upon activation, platelets flip phosphatidylserine (PS) to their outer membrane, which can be detected via lactadherin binding.³⁰ We found no difference in lactadherin signal between CD and TRT PLT at both storage time points; however, lactadherin signal increased 2–3 fold by D7 compared to D1 for both groups (Figure 2E). There was no difference in CD62P (marker of alpha granules) GMFI between CD and TRT on D1; by D7 there was a small yet statistically significant reduction in both CD and TRT compared to D1 (Figure 2F). There was no statistically significant difference between CD and TRT in expression of the activated conformation of CD41/CD61 (GPIIb/IIIb, binds fibrinogen), despite a 2-fold reduction in median GMFI values (Figure 2G). Notably, CD41/61 expression was greatly reduced in both groups by D7 (CD, p < .0001; TRT, p = .0008). There was no difference in CD63 (marker of dense granules) GMFI between CD and TRT on D1 or D7 (Figure 2H). These data highlight the phenotypic similarity of TRT and CD platelets over the course of storage.

We next assessed aggregation responses in TRT and CD PLT. Aggregation in response to ADP activation was similar between CD and TRT on D1 and D7, with both groups incapable of aggregating in response to ADP by D7 (Figure 3A–C). In contrast, we found a slight impairment in aggregation in response to collagen stimulation on D1 in TRT PLT compared to CD PLT (Figure 3D–F). On D1, the AUC was approximately 2-fold reduced in TRT PLT (p = .083; Figure 3E) and the % maximum amplitude was 2.5-fold reduced (p = .092; Figure 3F). Both CD and TRT PLT were minimally responsive to collagen stimulation by D7 of storage (Figure 3D–F).

FIGURE 3.

Stored platelet aggregation in response to ADP and Collagen agonists. (A) Aggregation over time in response to stimulation with ADP (10 μM, % per min), (B) area under the curve (AUC), and (C) percentage maximum aggregation (% Max Amp). (D) Aggregation over time in response to stimulation with collagen (5 μg/mL, % per min), (E) area under the curve (AUC), and (F) percentage maximum aggregation (% Max Amp). For A & D, data are visualized as mean + SEM; gray circles denote CD D1, gray squares denote CD D7, blue triangles denote TRT D1, and blue diamonds denote TRT D7. For B, C, E, F, data are visualized as median ± IQR, N = 10 per group; circles denote CD and triangles denote TRT; data were analyzed using two-way ANOVA (Fisher’s LSD test). Statistical significance between CD on D1 and D7 denoted as: ^##p < .01; ^####p < .0001. Statistical significance between TRT on D1 and D7 denoted as: ^{$ $}p < .01; ^{$ $ $ $}p < .0001.

There were little to no differences in the ability of CD and TRT PLT to support thrombin generation throughout storage duration (Figure 4). Curve analysis revealed no difference in thrombin generation kinetics between groups (Figure 4A), but with a slight reduction in maximal thrombin generation by TRT PLT on D1. However, this difference was not statistically significant (CD D1 peak 109.4 [89.31–132.8] vs. TRT D1 peak 94.21 [58.49–114.1], p = .1434; Figure 4B). Lagtime was similar between groups at D1, with a slightly shorter lagtime in TRT PLT on D7 compared to D1 (Figure 4C). Time to peak thrombin, that is, time taken to reach maximum thrombin generation, also stayed similar between groups and over time (Figure 4D). The AUC, also known as the ETP, was slightly elevated by D7 in both CD and TRT PLT compared to D1 (Figure 4E).

FIGURE 4.

Thrombin generation of CD and TRT platelets. (A) Thrombin generation curve over time, (B) maximum amount of thrombin generated (Peak Thrombin, nM), (C) time to thrombin generation (Lagtime, min), (D) time of maximum thrombin generation (Time to Peak Thrombin, min), and (E) endogenous thrombin potential (ETP, nM × min). For (A), data are visualized as mean + SEM; gray circles denote CD D1, gray squares denote CD D7, blue triangles denote TRT D1, and blue diamonds denote TRT D7. For (B–E), data are visualized as median ± IQR, N = 10 per group; circles denote CD and triangles denote TRT; data were analyzed using two-way ANOVA (Fisher’s LSD test). Statistical significance between CD on D1 and D7 denoted as: ^#p < .05. Statistical significance between TRT on D1 and D7 denoted as: ^$p < .05; ^{$ $}p < .01.

We next assessed the intrinsic global hemostatic capacity of CD and TRT PLT by measuring thrombus formation induced by collagen and tissue factor under conditions of physiological shear. TRT PLT appeared to have slightly reduced capacity to form occlusive thrombus on D1, as visualized in the right shift of the curve in Figure 5A compared to D1 CD PLT. However, there were no statistical differences between CD and TRT PLT on D1 in AUC (Figure 5B), occlusion start time (OST; Figure 5C), occlusion time (OT; Figure 5D), and growth time (GT; Figure 5E)—all metrics extrapolated from the curves in Figure 5A. Both D7 CD and TRT PLT were incapable of forming occlusive thrombus under arterial shear conditions as demonstrated by flat curves and close to zero values for all assay parameters.

FIGURE 5.

Hemostatic capacity of stored platelets under physiological flow conditions. (A) Total thrombus formation over time (kPa per min; total occlusion at 60 kPa denoted by dashed line), (B) area under the curve (AUC), (C) occlusion start time (OST, sec), (D) occlusion time (OT, sec), and (E) growth time (GT, sec). For (A), data are visualized as mean + SEM; gray circles denote CD D1, gray squares denote CD D7, blue triangles denote TRT D1, and blue diamonds denote TRT D7. For (B–E), data are visualized as median ± IQR, N = 10 per group; circles denote CD and triangles denote TRT; data were analyzed using two-way ANOVA (Fisher’s LSD test). Statistical significance between CD on D1 and D7 denoted as: ^####p < .0001. Statistical significance between TRT on D1 and D7 denoted as: ^{$ $}p < .01; ^{$ $ $}p < .001;^{$ $ $ $}p < .0001.

Due to the slight, but not statistically significant, impairments seen in both collagen-induced aggregation and thrombus formation assays, we posited perhaps TRT could be affecting the metabolic capacity of platelets responses to collagen activation, a bioenergetically demanding stimulus. We first compared the mitochondrial capacity of TRT and CD PLT using the Seahorse XFe24 platform (Figure 6). We found the mitochondrial capacity was similar between CD and TRT PLT at D1, and also similar between both groups at D7. However, there was reduced basal respiration in D7 PLT compared to D1 PLT, as expected (Figure 6A,B). D7 CD and TRT PLT had minimal responsiveness to FCCP stimulation, indicating little to no spare respiratory capacity. Direct comparison of D1 CD and TRT PLT found no differences in basal respiration (Figure 6C), maximal respiration (Figure 6D), spare respiratory capacity (Figure 6E), ATP production (Figure 6F), proton leak (Figure 6G), and non-mitochondrial respiration (Figure 6H). Comparisons of D7 TRT and CD PLT are detailed in Figure S3. In addition to mitochondrial assessment, we found no differences in key metabolic parameters of PLT products. Glucose levels were similar between CD and TRT PLT on D1, with a similar reduction over storage in both CD and TRT PLT (Figure 6I). As expected, lactate levels rose from D1 to D7 in both groups (Figure 6J). Bicarbonate (HCO₃₋) and oxygen levels were similar between both groups over time (Figure S4). Overall, TRT PLT did not exhibit reduced metabolic capacity compared to CD PLT.

FIGURE 6.

Mitochondrial function and metabolic activity in CD and TRT platelets. Oxygen consumption rate (OCR) over time on D1 and D7 of storage in (A) CD platelets and (B) TRT platelets. Injection timing of oligomycin (Oligo), FCCP, and rotenone and antimycin A (R/A) denoted by dotted lines. Data are visualized as mean ± SEM; gray circles denote CD D1, gray squares denote CD D7, blue triangles denote TRT D1, and blue diamonds denote TRT D7. (C) Basal respiration, (D) maximal respiration, (E) spare respiratory capacity, (F) ATP production, (G) proton leak, and (H) non-mitochondrial respiration of CD and TRT platelets on D1 of storage. (I) Glucose (mmol/L) and (J) lactate (mmol/L) levels in CD and TRT platelets on D1 and D7 of storage. For (C–J), data are visualized as median ± IQR, N = 10 per group; circles denote CD and triangles denote TRT. For (C–H), data were analyzed using a Mann–Whitney test, whereas for I & J, data were analyzed using two-way ANOVA (Fisher’s LSD test). Statistical significance between CD on D1 and D7 denoted as: ^####p < .0001. Statistical significance between TRT on D1 and D7 denoted as: ^{$ $ $ $}p < .0001.

4 |. DISCUSSION

The data presented herein highlight the similar hemostatic phenotype and function between PLT collected from CD and donors on TRT, providing much needed evidence in the case for using platelets from therapeutic phlebotomy of TRT donors. Evidence supporting the effects of testosterone on platelet function comes from in vivo and in vitro studies. Of note, in men with high risk of cardiovascular disease the use of TRT is associated with increased incidents of venous thromboembolism (VTE),³¹ a process in which platelets play a significant role.³² Platelets express functional androgen receptors that can be activated by elevated free testosterone in blood,³³ and supraphysiological doses of anabolic androgen steroids, such as testosterone, have been associated with increased platelet count, activity, thrombopoiesis, and aggregation.²⁰ While we observed elevated total and free testosterone levels in sera from TRT donors compared to controls (Figure S2), only three TRT donors had supraphysiological testosterone levels, and we found no correlation between testosterone level and platelet activity in TRT donors (Table S6). Moreover, our comprehensive analytical approach encompassed metrics associated with both primary (aggregate formation) and secondary hemostasis (thrombin generation) and demonstrated no reduction in intrinsic hemostatic function in PLT derived from TRT donors. Our data highlight how similar stored platelets collected from control and TRT donors are, and provide support for additional studies evaluating the safety of transfusing products with elevated testosterone levels.

This latter concept is of key importance, as there is a paucity of controlled trials examining the effects of supraphysiological testosterone administration via transfusion in various patient populations, to include those most likely to be affected by increasing amounts of testosterone (i.e., female recipients). Our data highlight a maximal amount of 400 pg/mL of free (bioactive) testosterone in the serum of donors (Figure S2). Assuming an average PLT volume of 300 mL, this would be approximately 120 ng of free testosterone that could be delivered during transfusion of a single platelet unit from a TRT donor. This amount of free testosterone is 0.04% of the estimated daily testosterone production in adult women (300 μg).³⁴ Furthermore, the liver will metabolize free testosterone relatively quickly into inactive metabolites which are then excreted via urine.³⁵ Of note, these assumptions are made contingent upon adult females and proper liver function. This suggests that these assumptions cannot hold in cases concerning liver failure or other androgen sensitive patients, such as female neonates requiring multiple transfusions.¹⁹ To this end, the likely effects of free testosterone in adult women receiving PLTs sourced from TRT donors are most likely minimal, yet further clinical studies are required to firmly evaluate these hypotheses.

While light transmission aggregometry has long been the gold standard for assessment of platelet function,^36,37 more recent developments have provided new tools for studying dynamic platelet function under shear regimes associated with physiological flow.³⁸ We have previously implemented the Total Thrombus Analysis System, or T-TAS, for studying intrinsic hemostatic function of stored PLT.³⁹ Herein, we again use the T-TAS to show that TRT does not affect the ability of PLT to form clot in settings of arterial shear (Figure 5). We also found no baseline difference in hemostatic capacity in whole blood samples from CD and TRT donors (Figure S5). Notably, D7 PLT from both CD and TRT are ineffective at forming an occlusive thrombus (Figure 5). D7 PLT from both CD and TRT have poor aggregation (Figure 3) and minimal metabolic responsiveness (Figure 6). Remarkably, the only functionality D7 PLT appear to have is that of supporting thrombin generation (Figure 4); however thrombin generation in D7 PLT is not significantly enhanced compared to D1 PLT, despite having 2–3 fold increased PS exposure (Figure 2E). Collectively, these data both demonstrate the limited functionality of platelets stored at RT for 7 days (as seen in other studies^39–44), as well as further advocate for the use of newer assays to measure dynamic platelet function which more fully reflect platelet function in vivo.

There are some limitations of our work which should be addressed in future expanded studies. First, PLT were generated from whole blood donations and processed in accordance with protocols used for generating whole-blood derived pooled PLT. As greater than 95% of PLT used in the US are single donor apheresis products,² the next iteration of this study should involve testing apheresis products derived from TRT donors. Notably, outside of the US, whole blood-derived pooled platelet concentrates make up a much larger fraction of the platelet inventory—closer to 30%–50%.⁴⁵ Therefore, another additional study could focus on pooling platelets derived from control and TRT donors, thereby reducing the testosterone concentration in the final pooled product all while maintaining hemostatic functionality.

Second, we observed significantly lower platelet count in products derived from TRT donors (Figure 1F). Current clinical practice highlights the importance of the corrected count increment (CCI) in determining platelet transfusion effectiveness. As TRT PLTs had lower counts, this highlights the possibility that these products may not be as effective due to a potential lower CCI. However, the PLADO trial demonstrated that there were no statistically significant differences in 4-h CCI between patient groups receiving low-, medium-, and high-dose PLTs (1.1 × 10¹¹, 2.2 × 10¹¹, transfusion, and 4.4 × 10¹¹ platelets per square meter per respectively), suggesting even a 75% reduction in platelet dose can still yield an effective transfusion per CCI values.⁴⁶ In addition, while CCI does provide information regarding the number of circulating platelets and thus efficacy of the act of transfusion, CCI does not in fact provide information on the quality, functionality, or hemostatic capacity of the transfused platelets. In support of this, Leitner et al. published a lack of correlation between CCI and hemostatic capacity (as measured by thromboelastography) in oncology patients receiving hematopoietic stem cell transplants.⁴⁷ Thus, while we observed lower counts in TRT donors, this may not be of functional significance post-transfusion, and requires further in vivo studies to determine any alterations in transfusion efficacy and functionality. Mechanistically, this reduced platelet number may be due to elevated HCT in TRT donors (Figure 1D) which could affect RBC sedimentation during centrifugation,^48,49 and in turn, affect platelet yields. This may also be attributable to enhanced platelet:RBC interactions; a recent study showed how platelets can mediate clearance of senescent RBC by forming pro-phagocytic platelet-cell complexes to remove old RBC from circulation.⁵⁰ TRT is known to affect RBC membrane integrity,⁵¹ a key feature of RBC senescence, and to increase HCT.^11–13 Taken together, this could promote a greater likelihood of platelet:RBC complex formation, preventing proper platelet isolation. Additional studies are warranted to understand platelet:RBC interactions in the setting of TRT and how this affects downstream processing of PLT derived from TRT donors.

In conclusion, we have shown here, with extensive characterization, that there are no significant differences in the hemostatic quality of platelets derived from TRT donors. Despite TRT donors only representing 1.6% of the total donor population at our blood center²⁷ (estimates based on RBC product collections), this could represent an additional source of platelet donors to help alleviate critical shortages in platelet inventories,²⁷ especially as platelet donation is allowed every 2 weeks in contrast to the 8 weeks required between whole blood donations. While not a large number of potential products, any additionally available products may help reduce the burden on platelet inventories, in addition to the other mechanisms outlined in the introduction. Our study supports the need for additional pre-clinical and clinical studies to evaluate the safety of PLT containing elevated levels of free testosterone, and to advocate for the expansion of FDA donor criteria to allow PLT from TRT donors be distributed for transfusion.

Abbreviations:

ADP

adenosine diphosphate

ARUP

Associated Regional and University Pathologists, Inc.

ATP

adenosine triphosphate

AUC

area under the curve

CAT

calibrated automated thrombogram

CBC

complete blood count

CCI

corrected count increment

CD

control donor

CD41

integrin αIIb

platelet glycoprotein IIb

CD41/61

integrin αIIb β3

platelet glycoprotein IIb/IIIa

CD42b

platelet glycoprotein Ibα

CD49b

integrin α2

CD62P

P-selectin

COL

collagen

D1

day 1 of storage

D7

day 7 of storage

EDTA

ethylenediaminetetraacetic acid

ETP

endogenous thrombin potential

FCCP

carbonyl cyanide-4 phenylhydrazone

FDA

Food & Drug Administration

FITC

fluorescein isothiocyanate

GMFI

geometric mean fluorescence intensity

GPVI

platelet glycoprotein VI

GT

growth time

HCT

hematocrit

HGB

hemoglobin

IQR

interquartile range

MCV

mean corpuscular volume

MPV

mean platelet volume

OCR

oxygen consumption rate

OST

occlusion start time

OT

occlusion time

PLT

platelet product

PRP

platelet-rich plasma

RBC

red blood cell

RT

room temperature

SEM

standard error of the mean

SHBG

sex hormone binding globulin

TRT

Testosterone Replacement Therapy

T-TAS

total thrombus analysis system

TXA2

thromboxane A2

US

United States

VTE

venous thromboembolism